
\begin{figure}[H]
\centering
\caption{RD First Stages of 2018 and 2019 Interventions}
\label{fig:firststagesRD18-19}
%[1,1]
	\begin{subfigure}[t]{\sizeSubfig}
		\centering
		\caption{First Stage Total Pop-Ups-- 2018}
		%\label{ }
		\includegraphics[width=\widthPng]{figuresCL/firstStages/2018_P_fs_totalPopup_allwidth.png}
	\end{subfigure}%
%[1,2]
	\begin{subfigure}[t]{\sizeSubfig}
		\centering
		\caption{First Stage Reminder SMS-- 2018}
		%\label{ }
		\includegraphics[width=\widthPng]{figuresCL/firstStages/2018_P_fs_SMSReminder_allwidth.png}
	\end{subfigure}
	
	\vspace{\vgap}
	
	\begin{subfigure}[t]{\sizeSubfigWide}
		\centering
		\caption{First Stage Total Pop-Ups--2019}
		%\label{ }
		\includegraphics[width=\widthPng]{figuresCL/firstStages/2019_P_fs_totalPopup_allwidth.png}
	\end{subfigure}	
\floatfoot{\footnotesize{\textit{Notes.} Warnings interventions in 2018 and 2019, by position relative to risk score prediction. Panel A: Count of platform pop-ups received in 2018 intervention. Panel B: count of reminder SMS messages received in 2018 intervention. Panel C: count of platform pop-ups received in 2019 intervention. 2019 intervention did not include SMS reminders. Points are centered means of 50 quantile-spaced bins of the support of the predicted placement risk $\in [0.02 ; 0.98]$. Solid line shows the quadratic fit. Reported coefficients and standard errors are from local linear specifications using $+-0.1$ bandwidth.  See section \ref{sec:intdes} for intervention details.  } }
\end{figure}
